Towards Content Provider Aware Recommender Systems: A Simulation Study on the Interplay between User and Provider Utilities

Most existing recommender systems focus primarily on matching users (content consumers) to content which maximizes user satisfaction on the platform. It is increasingly obvious, however, that content providers have a critical influence on user satisfaction through content creation, largely determining the content pool available for recommendation. A natural question thus arises: can we design recommenders taking into account the long-term utility of both users and content providers? By doing so, we hope to sustain more content providers and a more diverse content pool for long-term user satisfaction. Understanding the full impact of recommendations on both user and content provider groups is challenging. This paper aims to serve as a research investigation of one approach toward building a content provider aware recommender, and evaluating its impact in a simulated setup.